Natural Muscle September 2012
Dortha Chuang
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Molecular screens comparing different disease states to identify candidate genes rely on the availability of fast, reliable and multiplexable systems to interrogate genes of interest. CRISPR/Cas9-based reverse genetics is a promising method to eventually achieve this. However, such methods are sorely lacking for multi-nucleated muscle fibers, since highly efficient nuclei editing is a requisite to robustly inactive candidate genes. Here, we couple Cre-mediated skeletal muscle fiber-specific Cas9 expression with myotropic adeno-associated virus-mediated sgRNA delivery to establish a system for highly effective somatic gene deletions in mice. Using well-characterized genes, we show that local or systemic inactivation of these genes copy the phenotype of traditional gene-knockout mouse models. Thus, this proof-of-principle study establishes a method to unravel the function of individual genes or entire signaling pathways in adult skeletal muscle fibers without the cumbersome requirement of generating knockout mice.
With the advent of -omics technologies that allow to correlate molecular signatures with specific disease states of cells or tissues, there is an increasing need for methods to interrogate the function of genes and pathways. Traditionally, forward and reverse genetics using targeted mutagenesis in combination with transgenesis has been used. More recently, clustered regularly interspaced short palindromic repeats (CRISPR)-mediated genome editing has become the method of choice for gene engineering in many species and tissues1.
When it comes to skeletal muscle tissue, studying gene function in vivo is particularly challenging. Skeletal muscle is one of the largest organs constituting up to 50% of the mammalian body mass2. The size and the fact that muscle fibers, which are the functional contractile units of skeletal muscle, form a syncytium with hundreds of myonuclei in a common cytosol, represent a substantial challenge for somatic gene inactivation. Therefore, the method of choice for functional gene interrogation studies in muscle remains transgenic mice generated via the Cre-loxP system. However, generation of transgenic mice requires extensive breeding, making functional interrogation of multiple genes cumbersome and time consuming.
Effective methods for somatic gene perturbation would offer huge advantages for screening multiple muscle gene candidates. While RNA interference, which can silence a target gene by introducing short hairpin (sh) RNAs3, can acutely silence gene expression in muscle fibers4,5, prolonged elimination of a gene product requires sustained, high expression of the shRNA. The introduction of viruses, in particular adeno-associated viruses (AAV), as vehicles for delivering shRNAs, opened the possibility of systemic administration6. However, due to the lack of tissue-specific control of shRNA expression, gene silencing usually occurs in all transduced cells. While next-generation AAV capsids with designed tropism towards skeletal muscle tissue7,8,9 may improve the off-tissue targeting, all of them also target myocytes in the heart. Another challenge for somatic gene targeting of muscle fibers is the overall heterogeneity of the tissue. Almost half of the nuclei in skeletal muscle derive from non-fiber cells, such as muscle stem cells (MuSC), endothelial cells, fibro-adipogenic precursors (FAPs), Schwann cells or tenocytes10 and perturbation of their function often affects muscle fibers as well. Therefore, for rapid functional gene interrogation in skeletal muscle fibers, an efficient, multiplexable, and muscle fiber-specific gene editing approach is sorely needed.
Here we establish a versatile tool for local and systemic skeletal muscle fiber-specific gene knockout. This tool couples the advantages of CRISPR with recently developed, highly efficacious, AAV9-derived viral capsids by using (i) mice engineered to constitutively or inducibly express Cas9 in skeletal muscle fibers and (ii) delivering single guide (sg) RNAs with the myotropic AAVMYO8. By targeting key genes, we demonstrate that this system is capable of potently altering signaling pathways, destroying neuromuscular junctions, and stimulating muscle hypertrophy without needing to generate germline gene-of-interest deletions.
To test whether our system could also drive gain of muscle function, we next targeted myostatin (GDF-8), a TGF-β family protein secreted by skeletal muscle that acts as an inhibitor of muscle size28. Deletion of Mstn in mice results in robust muscle hypertrophy29 and naturally occurring Mstn null-mutants cause hypermuscularity in many species, including cows and humans30,31. Myostatin signals through a combination of type-2 and type-1 receptors. This signaling pathway is also activated by several other ligands, including activin. The two ligand-binding receptors are activin A receptor type-2/IIA (ACVR2A or ACTRIIA) and type-2/IIB (AAVR2B or ACTRIIB). The activin A type-2 receptors are partially redundant as targeting both receptors elicits stronger muscle hypertrophy than deletion of each receptor individually32. Upon ligand-binding, the type-2 receptors form a complex with type-1 activin A receptor-like kinase-4 (ALK4) and ALK5, which are also partially redundant, to trigger intracellular signaling.
This article presents a rapid and highly efficient tool to investigate the function of single or multiple genes in adult skeletal muscle fibers. Feasibility and efficiency of the system is demonstrated by knocking out essential genes for the integrity of the NMJ and for skeletal muscle fiber growth.
We show that high, long-term Cas9 expression in skeletal muscle fibers does not affect muscle size or function. Others have used AAV to deliver Cre to LSL-Cas9KI mice to excise the stop cassette and drive Cas9 expression11,36. While this approach allows for the use of Cas9-GFP as a transfection marker and reduces any potential side effects of prolonged Cas9 expression, AAV-mediated delivery of Cre would also lead to Cas9 expression in any AAV-targeted tissues, including the heart and liver. To our best knowledge, highly specific AAV-compatible promoters for skeletal muscle fibers do not currently exist. As such, our AAVMYO-CRISPR/Cas9mKI strategy represents a major advancement for somatic gene perturbation of mouse skeletal muscle fibers.
While CRISPR/Cas9 systems for somatic gene deletion have been described for some tissues, including brain and liver11,37, such a versatile tool has so far been missing for skeletal muscle fibers. Previous work has demonstrated successful somatic gene editing using CRISPR in muscle stem cells (MuSCs) although with rather low efficiency38,39. While editing efficiency can be increased by sorting MuSCs based on a fluorescent transfection marker, this is not possible for multi-nucleated skeletal muscle fibers. Thus, the successful depletion of a gene by CRISPR in muscle fibers is only possible when indels are generated in both alleles in the majority of myonuclei. Such high efficiencies are not required in CRISPR/Cas9-mediated editing approaches that aim to correct gene mutations causing muscular dystrophies38,40,41,42,43,44. In these experiments, correcting the mutation in a subset of myonuclei and in one allele is sufficient as the corrected protein will distribute in a large part of the muscle fiber cytoplasm.
As a functional proof-of-concept, we perturbed MuSK function, which is essential for NMJ formation and maintenance21. Musk expression in adult mice is confined to sub-synaptic nuclei, which lay directly underneath the NMJ. Sub-synaptic Musk expression is based on local, NMJ-derived signals that overwrite activity-mediated transcription suppression in non-synaptic myonuclei21. Denervation and hence loss of electrical activity results in Musk re-expression in non-synaptic myonuclei. Thus, efficient editing in myonuclear DNA is important to abrogate Musk expression in in muscle fibers. At the highest AAVMYO dose, Musk transcripts were reduced up to 98%. The reason for this strong loss of Musk expression is based on the use of multiple sgRNAs that introduce large deletions between different target sites, which are likely to de-stabilize mRNA. Indeed, PCR amplification of cDNA derived from polyA-positive mRNA showed the presence of many truncated products between the individual sgRNA-target sites (Fig. S11). We also detected many short indels for each sgRNA in Musk (Fig. S10), which may result in frameshifts and the occurrence of premature termination codons that cause nonsense-mediated mRNA decay. A strong reduction of transcript levels was also observed for Acvr2a and Acvr2b using multiple sgRNAs. We also tested whether the use of multiple sgRNAs would increase off-target editing. Out of the 22 primary off-target sites tested, only two exhibited rather high DNA editing. However, both loci are in intronic or intergenic regions and hence should not impair function of nearby genes. Moreover, the somatic gene perturbation studies shown here did not last more than a few weeks, which makes it unlikely that off-target editing might be responsible for any of the observed effects. We hence advise to not use sgRNAs for which off-target sites are located in exons and to confirm the results of perturbation experiments for genes with unknown function with a second, independent set of sgRNAs targeting the same gene.
We also demonstrate efficacious, simultaneous inactivation of multiple genes (Acvr2a and Acvr2b) with the AAVMYO-CRISPR/Cas9 system, opening the possibility of studying several genes or signaling pathways concurrently. Although our experiments targeting Prkca indicate that one sgRNA can be sufficient to eliminate a gene, testing each sgRNA in vitro prior to in vivo application is laborious. We find that the targeting efficiency of individual sgRNAs varies in myonuclei between approximately 20% to 60% (Figs. S10 and S13) and that the use of multiple sgRNAs generates bigger deletions (Fig. S11). Hence, we suggest targeting each gene with several different sgRNAs, minimizing the risk of insufficient protein loss. Since one AAV has sufficient packaging capacity for at least 7 sgRNAs, we anticipate that up to three genes could be silenced with one AAV. By delivering two AAVs (as done here for Acvr2a and Acvr2b), up to six independent genes could be silenced simultaneously, allowing interrogation of entire signaling pathways, specifically in skeletal muscle fibers.
b1e95dc632